The Maui High Performance Computing Center (MHPCC) was established in 1993 within the Department of Defense's (DoD) High Performance Computing Modernization Program (HPCMP). In 2014, Congress asked DoD to cut about $45 million from the HPCMP budget. This report evaluates the capabilities, challenges, and future options for the MHPCC, providing important context about the center as decisionmakers work toward resolving its budget gap.
Assessing the Need for Supercomputing Resources Within the Pacific Area of Responsibility
- Which of MHPCC's capabilities are used most often by customers in the Pacific area of responsibility (AOR), notably U.S. Pacific Command?
- And, specifically, is there a demand for a supercomputing resource that is located within the AOR?
- What recommendations should AFRL consider when making decisions about the future of MHPCC?
This report evaluates the capabilities, challenges, and future options for the Maui High Performance Computing Center (MHPCC). Located in Kihei on the island of Maui, Hawaii, the MHPCC was established in 1993 as a center within the Department of Defense's (DoD) High Performance Computing Modernization Program (HPCMP). As one of five HPCMP centers, MHPCC's primary mission is to provide computing cycles and other high performance computing capabilities to DoD's research, development, test and evaluation community. In the fall 2014, Congress asked DoD to consider ways to cut about $45 million from the HPCMP so future fiscal years could be supported entirely by the Army's programmed budget of $183 million. The findings within this report are designed to provide decisionmakers with important context about potential future missions and use cases for MHPCC as Congress, the Army, and the HPCMP work toward a solution for resolving the budget gap.
- There is a consistent lack of understanding among the customer base as to what problems are suitable for a supercomputer or high performance computing assets, in general.
- While several entities within PACOM offered a stated preference for a supercomputing facility within the geographic region, we were unable to identify any entity within the Pacific AOR with a quantitative requirement for real time supercomputing. In the absence of such a requirement, we did gather anecdotal evidence for the value that PACOM users receive by having HPC expertise located within the AOR.
- PACOM users value MHPCC's ability to stand up and support custom servers. These users also value MHPCC's resident expertise in programming for high performance computing architectures.
- MHPCC and AFRL's Maui optical observatory both benefit from being co-located.
- Be explicit about the different capabilities at MHPCC. Potential users need to be aware of what MHPCC has to offer, what is available from other entities within the HPCMP enterprise, and, if appropriate, what other services are available outside the HPCMP.
- MHPCC should market itself as offering three separate products: a 12,096-node supercomputer; a facility with the expertise and infrastructure to host standalone HPC assets; and a research staff with the capacity to develop creative solutions to operational problems. Presenting MHPCC as a center that offers separate products will allow for a more-tailored approach to solving customer problems.
- Provide PACOM users with a framework for matching HPC assets to the desired workflow. HPC is inherently technical, and PACOM customers need simple frameworks to understand how their problem fits into the greater context.
- Employ strategic and technical thought leaders to enhance Pacific AOR customer exploitation of MHPCC capabilities. Strategists need to identify areas where Pacific AOR users might benefit from HPC capabilities, and the strategists need to work with the subject-matter experts to help them develop future requirements for HPC assets.
- Rebuild trust and understanding with the HPCMP leadership. Reestablishing trust always takes time, but MHPCC management can take some initial steps by setting up consistent communication with HPCMP leadership.
- Recognize that aspects of the computing enterprise are changing. As computing cycles become more of a commodity, it will be important to develop a value proposition that focuses more on MHPCC's human capital and is not tied to hardware alone.
Table of Contents